AI security identification method and system based on BIM and intelligent construction site

Abstract

The invention provides an AI security identification method and system based on BIM and an intelligent construction site, and the method comprises the steps: introducing BIM model security index estimation data to generate BIM model security index prior data through effectively processing part of missing security identification data, especially for BIM model security index annotation data with feature missing, and improving the security identification accuracy. The method not only improves the robustness and accuracy of the AI security optimization network in processing incomplete data, but also significantly optimizes the network parameter learning process, so that the generated AI security optimization network can efficiently and accurately output the BIM model data with the target security identification index, thereby enhancing the application efficiency of the BIM model in the security identification aspect, and improving the security identification efficiency of the AI security optimization network. And data support is provided for safety management of the intelligent construction site.

Description

Technical Field

[0001] The present application relates to the technical field of smart construction sites, and more specifically, to an AI safety identification method and system based on BIM and smart construction sites. Background Art

[0002] With the widespread application of Building Information Modeling (BIM) technology in the construction industry, BIM models have become an indispensable tool in the process of building design, construction and management. BIM models not only contain the geometric information of buildings, but also cover a large amount of safety indicator data, which is crucial to ensure the safety of building construction. However, in actual applications, due to various reasons (such as incomplete data collection, incorrect annotation, etc.), the safety indicator data in the BIM model is often partially missing, which brings great challenges to subsequent safety management and risk assessment.

[0003] Traditional solutions mainly rely on manual inspection and empirical judgment, which is not only inefficient but also easily affected by human factors, resulting in low accuracy and reliability of safety identification. In recent years, with the rapid development of artificial intelligence (AI) technology, it has become a trend to apply AI technology to safety identification of BIM models. However, existing AI safety identification methods often fail to achieve satisfactory results when processing partially missing safety identification data. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present application, an embodiment of the present application provides an AI safety identification method based on BIM and smart construction sites, the method comprising: Obtaining partially missing safety identification data of a sample BIM model sequence learned by an AI safety optimization network, wherein the partially missing safety identification data includes X BIM model safety indicator annotation data, each BIM model safety indicator annotation data includes annotated safety indicator parameters of a BIM model for Y safety indicators, and at least one BIM model safety indicator annotation data with missing features exists among the X BIM model safety indicator annotation data, and at least one labeled safety indicator parameter of a safety indicator is missing among the labeled safety indicator parameters of the Y safety indicators included in the BIM model safety indicator annotation data with missing features; For each sample BIM model, if the BIM model safety index annotation data corresponding to the sample BIM model in the partially missing safety identification data is feature-missing, then based on the BIM model safety index estimation data and the BIM model safety index annotation data of the sample BIM model, the BIM model safety index prior data of the sample BIM model is generated, and the BIM model safety index estimation data of the sample BIM model includes: estimated safety index parameters of the sample BIM model for at least one safety index generated by estimation; Based on the BIM model safety index prior data of the sample BIM model, network parameter learning is performed on the AI ​​safety optimization network to generate an AI safety optimization network that has completed network parameter learning. The AI ​​safety optimization network that has completed network parameter learning is used to optimize and generate BIM model data with target safety identification indicators.

[0005] On the other hand, an embodiment of the present application also provides a smart construction site service system, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, the embodiment of the present application effectively processes some missing safety identification data, especially for BIM model safety indicator annotation data with missing features, and introduces BIM model safety indicator estimation data to generate BIM model safety indicator prior data. This not only improves the robustness and accuracy of the AI ​​safety optimization network when processing incomplete data, but also significantly optimizes the network parameter learning process, so that the generated AI safety optimization network can efficiently and accurately output BIM model data with target safety identification indicators, thereby enhancing the application efficiency of the BIM model in safety identification and providing data support for the safety management of smart construction sites. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the execution flow of the AI ​​safety identification method based on BIM and smart construction sites provided in an embodiment of the present application.

[0008] Figure 2 It is a schematic diagram of the hardware architecture of the smart construction site service system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0009] The present application will be described in detail below with reference to the accompanying drawings. Figure 1This is a flow chart of an AI safety identification method based on BIM and smart construction sites provided by an embodiment of the present application. The AI ​​safety identification method based on BIM and smart construction sites is introduced in detail below.

[0010] Step S110, obtaining partially missing safety identification data of a sample BIM model sequence learned by the AI ​​safety optimization network, wherein the partially missing safety identification data includes X BIM model safety indicator labeling data, each BIM model safety indicator labeling data includes a labeled safety indicator parameter of a BIM model for Y safety indicators, and there is at least one BIM model safety indicator labeling data with feature missing among the X BIM model safety indicator labeling data, and there is a missing labeled safety indicator parameter of at least one safety indicator among the labeled safety indicator parameters of the Y safety indicators included in the BIM model safety indicator labeling data with feature missing.

[0011] In detail, the AI ​​safety optimization network is a network model based on artificial intelligence (AI) technology, which is specifically used to optimize the safety identification indicators of BIM models. It can automatically adjust model parameters by learning and analyzing a large amount of BIM model data to generate a BIM model that meets specific safety requirements.

[0012] The sample BIM model sequence is a series of BIM models used to train and test the AI ​​safety optimization network, which typically covers different design stages, design teams, and building structure parts to ensure that the network can learn a wide range of safety knowledge and features.

[0013] The partially missing safety identification data refers to the situation where some of the collected BIM model safety identification data is incomplete or missing. These data may not contain all the necessary safety indicator parameters due to various reasons (such as incomplete data collection, data transmission errors, etc.).

[0014] The BIM model safety indicator annotation data is data for annotating each safety indicator in the BIM model. Each BIM model will have a set of corresponding safety indicator annotation data, which is used to describe the safety characteristics and performance of the model. BIM model safety indicator annotation data with missing features refers to the situation where there is a missing annotation parameter of at least one safety indicator in the BIM model safety indicator annotation data. This missing may be caused by problems in the data collection or transmission process.

[0015] For example, in this embodiment, in the process of building a BIM (Building Information Model) safety management system for a large-scale construction project, this embodiment manages the data processing and model optimization of the entire process.

[0016] For example, in this construction project, there are multiple different types of building structures, such as the main structure, ancillary facilities, etc. Each part has a corresponding BIM model. These BIM models cover a wealth of information, including safety indicators in multiple aspects such as structural safety, fire safety, and electrical safety.

[0017] This embodiment has collected multiple sample BIM model sequences from different design stages and different design teams. For example, there is a set of sequences containing 10 sample BIM models from the preliminary design stage, and another set of sequences containing 8 sample BIM models from the detailed design stage, etc.

[0018] For each sample BIM model sequence, the server obtains its complete safety identification data. Taking the 10 sample BIM model sequences in the preliminary design stage as an example, the complete safety identification data of each sample BIM model may include annotated safety index parameters for 5 safety indicators (assuming Y=5, which are structural stability, fire protection level, electrical grounding, rationality of evacuation channels, and ventilation safety).

[0019] For example, for the safety indicator of structural stability, the marked safety indicator parameter may be the safety factor calculated according to specific structural mechanics; for the fire protection level, it may be the level value marked according to the building fire protection code (such as level 1, level 2, etc.).

[0020] In this embodiment, each BIM model can be used as a unit to merge the complete safety identification data of multiple sample BIM model sequences. In the merging process, data may be missing.

[0021] Assume that the complete safety identification data from three different sample BIM model sequences are merged, and a total of 20 BIM model safety identification data are obtained (i.e., X=20). In this process, some BIM model safety indicator annotation data may have missing features due to incomplete data collection on specific safety indicators of some models or data transmission errors.

[0022] For example, for one of the BIM models, due to a measurement device failure, the safety indicator parameters for the electrical grounding were not accurately annotated, resulting in BIM model safety indicator annotation data with missing features. Among the 20 BIM models with partially missing safety identification data, there may be 5 BIM model safety indicator annotation data with missing features.

[0023] Step S120: For each sample BIM model, if the BIM model safety indicator annotation data corresponding to the sample BIM model in the partially missing safety identification data is feature missing, then based on the BIM model safety indicator estimation data and the BIM model safety indicator annotation data of the sample BIM model, generate the BIM model safety indicator prior data of the sample BIM model, and the BIM model safety indicator estimation data of the sample BIM model includes: estimated safety indicator parameters of the sample BIM model for at least one safety indicator generated by estimation.

[0024] In detail, the BIM model safety index estimation data is estimated data generated by a special BIM model safety index estimation network. When there are missing features in the BIM model safety index annotation data, these data can be used to supplement or replace the missing annotation parameters. The BIM model safety index prior data is a priori information generated based on the BIM model safety index annotation data and estimation data, and is used to provide additional guidance and constraints in the learning process of the AI ​​safety optimization network to help the network more accurately generate a BIM model that meets safety requirements.

[0025] In this embodiment, a special BIM model safety index estimation network is pre-configured to generate BIM model safety index estimation data for the sample BIM model. For BIM models with missing features, this network will perform estimation based on other relevant information of the model.

[0026] For example, for a BIM model with missing features in ventilation safety indicators, the BIM model safety indicator estimation network will estimate the estimated safety indicator parameters of ventilation safety based on the model's building space layout, occupant density and other related information. It is assumed that this estimated safety indicator parameter is an evaluation value of ventilation effect obtained by learning a large number of similar building models based on a machine learning algorithm.

[0027] Taking a sample BIM model as an example, assume that the safety indicator parameter marked on the safety indicator of rationality of the evacuation channel of this sample BIM model is missing.

[0028] First, the server will check whether the annotated safety index parameters of the rationality of the evacuation channel in the BIM model safety index annotated data of the sample BIM model are missing. Since they are indeed missing, the server will obtain the estimated safety index parameters of the rationality of the evacuation channel from the BIM model safety index estimation data of the sample BIM model as the safety index prior parameters of this safety index.

[0029] For other safety indicators in this sample BIM model that do not have missing annotated safety indicator parameters, such as structural stability, the server will directly output the annotated safety indicator parameters of structural stability as safety indicator prior parameters of this safety indicator.

[0030] Finally, the server generates the BIM model safety index prior data of this sample BIM model based on the safety index prior parameters of these Y safety indicators (assuming Y=5).

[0031] Step S130, based on the BIM model safety index prior data of the sample BIM model, network parameter learning is performed on the AI ​​safety optimization network to generate an AI safety optimization network that has completed network parameter learning. The AI ​​safety optimization network that has completed network parameter learning is used to optimize and generate BIM model data with target safety identification indicators.

[0032] In this embodiment, the server generates an iterative BIM model corresponding to the sample BIM model based on the sample BIM model and the BIM model safety index prior data of the sample BIM model according to the AI ​​safety optimization network.

[0033] Among them, the AI ​​safety optimization network includes a first optimization branch and a second optimization branch. The first optimization branch is used to generate a hidden layer expression vector based on the sample BIM model and the BIM model safety index prior data of the sample BIM model. For example, for a sample BIM model, the first optimization branch generates a hidden layer expression vector based on the structural characteristics of the model, safety index prior data and other information, through calculation and conversion of hidden layer neurons. The average data and morphological vector of the distribution model parameters followed by this hidden layer expression vector are determined by the network parameter information of the first optimization branch, the sample BIM model and the BIM model safety index prior data of the sample BIM model.

[0034] Then, the second optimization branch generates an iterative BIM model based on the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model. The average data and morphological vector of the distribution model parameters followed by the iterative BIM model are determined by the network parameter information of the second optimization branch, the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model.

[0035] Assume that for a sample BIM model, the annotated safety indicator parameters of the structural stability annotated safety indicator in the BIM model safety indicator annotated data are complete, and the estimated safety indicator parameters of the structural stability are also included in the BIM model safety indicator estimation data. The server calculates the fit evaluation cost parameter based on these two parameters. For example, if the difference between the estimated safety indicator parameter and the annotated safety indicator parameter is large, the fit evaluation cost parameter will be high, which reflects the estimated performance indicator of the BIM model safety indicator estimation data.

[0036] Then, the server determines the first cost parameter based on the iterative BIM model and the sample BIM model. For example, if the building dimensions in the iterative BIM model are significantly different from those in the sample BIM model (this is an explicit feature), then the first cost parameter will reflect the explicit feature distance between the iterative BIM model and the sample BIM model, and the value will be larger.

[0037] Assuming that the BIM model safety index prior data of the sample BIM model has been generated, the server determines the second cost parameter and the third cost parameter based on the BIM model safety index prior data, the collaborative parameter array of the BIM model safety index prior data, the average data of the BIM model safety index prior data, and the average data of the BIM model safety index estimation data.

[0038] For example, for the annotated safety indicator (such as fire rating), if the overall characteristic distance between its safety indicator prior parameters and the BIM model safety indicator prior data is large, the second cost parameter will reflect this situation. For other safety indicators in Y safety indicators except the annotated safety indicator (such as electrical grounding), if the characteristic distance between its safety indicator prior parameters and the BIM model safety indicator prior data is large, the third cost parameter will reflect this difference.

[0039] Next, the server obtains the hidden layer expression vector generated for the sample BIM model from the hidden network layer of the AI ​​security optimization network. This hidden layer expression vector is the result of the sample BIM model being calculated and transformed by the hidden layer neurons. For example, this hidden layer expression vector may contain the representation of the sample BIM model in different abstract feature spaces.

[0040] Then the mean and shape vector in the distribution model parameters followed by the hidden layer expression vector are determined. The mean reflects the central position of the hidden layer expression vector in the current state distribution, and the shape vector is related to the distribution shape.

[0041] A hidden layer expression vector sample set is selected based on the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector, and a pre-built hypothetical distribution model is obtained (this hypothetical distribution model represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state distribution).

[0042] The KL divergence is used as a measure of the difference between the distribution model parameters of the hidden layer expression vectors in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model. According to the mean and covariance matrix of the distribution model parameters of the hidden layer expression vectors, and the covariance matrix associated with the mean and morphological vector corresponding to the assumed distribution model, the relevant parameters for calculating the fourth cost parameter are determined.

[0043] Based on the calculation method and related parameters of KL divergence, the fourth cost parameter is calculated by combining the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector and the difference between the mean and covariance matrix related to the morphological vector corresponding to the assumed distribution model. For example, if the distribution of the hidden layer expression vector is significantly different from the assumed distribution model, the fourth cost parameter will be larger. Finally, the calculated fourth cost parameter is divided by the reference value calculated based on the prior sample BIM model for normalization to generate the normalized fourth cost parameter.

[0044] Thus, the server determines the probability model optimization cost parameter based on the first cost parameter, the second cost parameter, the third cost parameter and the fourth cost parameter. The probability model optimization cost parameter reflects the characteristic distance between the iterative BIM model generated based on the BIM model safety index prior data and the sample BIM model.

[0045] Finally, the server evaluates the cost parameters based on the fit and optimizes the cost parameters based on the probability model to determine the training cost parameters.

[0046] In this embodiment, the server optimizes the network parameters of the AI ​​safety optimization network based on the training cost parameters. For example, if the training cost parameters are high, it means that the performance of the current AI safety optimization network needs to be improved, and the server will adjust the weights, biases and other parameters in the network to reduce the training cost parameters. After multiple iterations of optimization, an AI safety optimization network that completes network parameter learning is finally generated. This AI safety optimization network that has completed network parameter learning can be used to optimize the generation of BIM model data with target safety identification indicators. For example, in subsequent construction projects, if it is necessary to generate a BIM model that meets specific structural safety, fire safety and other target safety identification indicators, this optimized AI safety optimization network can be used for optimized generation.

[0047] If the number of BIM models in the partially missing safety identification data is greater than the threshold number (assuming the threshold number is 15). For example, in the partially missing safety identification data of the 20 BIM models mentioned above, the BIM model safety index estimation network will call the first deep learning network that has not yet performed network parameter pre-learning. In addition, during the network parameter learning process of the AI ​​safety optimization network, the network parameters of the first deep learning network are optimized based on the loss function value of the BIM model safety index annotation data of the sample BIM model and the BIM model safety index estimation data of the sample BIM model.

[0048] If the number of BIM models in the partially missing safety identification data is not greater than the threshold number (assuming that there are only 10 BIM models in the partially missing safety identification data), the BIM model safety index estimation network will call the second deep learning network after pre-learning the network parameters. This second deep learning network is an artificial intelligence network with BIM model safety index estimation performance generated after pre-learning the network parameters using the complete safety identification data of the sample BIM model sequence. The complete safety identification data includes a variety of sample BIM models and corresponding safety index parameters.

[0049] Based on the above steps, the embodiment of the present application effectively processes some missing safety identification data, especially for BIM model safety indicator annotation data with missing features, and introduces BIM model safety indicator estimation data to generate BIM model safety indicator prior data. This not only improves the robustness and accuracy of the AI ​​safety optimization network when processing incomplete data, but also significantly optimizes the network parameter learning process, so that the generated AI safety optimization network can efficiently and accurately output BIM model data with target safety identification indicators, thereby enhancing the application efficiency of the BIM model in safety identification and providing data support for the safety management of smart construction sites.

[0050] In a possible implementation, step S120 includes: Step S121: For each of the Y safety indicators, if the annotated safety indicator parameters of the safety indicator covered in the BIM model safety indicator annotation data of the sample BIM model are missing, the estimated safety indicator parameters of the safety indicator are obtained from the BIM model safety indicator estimation data of the sample BIM model as safety indicator prior parameters of the safety indicator.

[0051] Step S122: if there is no missing annotated safety indicator parameters of the safety indicator covered in the BIM model safety indicator annotation data of the sample BIM model, the annotated safety indicator parameters of the safety indicator are output as safety indicator priori parameters of the safety indicator.

[0052] Step S123: generating BIM model safety index prior data of the sample BIM model based on the safety index prior parameters of the Y safety indexes.

[0053] In this embodiment, in the aforementioned large-scale construction project, the server continuously processes and optimizes the safety indicators of the BIM model. Taking a project containing multiple different building structure parts (such as the main structure, ancillary facilities, etc.) as an example, each part has a corresponding BIM model, and these BIM models contain safety indicator information in various aspects such as structural safety, fire safety, and electrical safety.

[0054] Assume that we focus on a specific sample BIM model, which represents an important functional area in a building, such as an office area BIM model that contains many human activity spaces and complex equipment. Set Y=5, that is, there are 5 safety indicators, namely structural stability, fire protection level, good electrical grounding, reasonable evacuation channel, and ventilation safety.

[0055] For the structural stability index, the server first checks the annotated safety index parameters of the structural stability in the BIM model safety index annotation data of the sample BIM model.

[0056] In this building project, the annotated safety index parameters of the structural stability are obtained through detailed structural mechanics analysis, such as the safety factor of the structure under different load combinations. Assume that the server finds that the annotated safety index parameters of the sample BIM model are complete and there are no missing items.

[0057] Then, according to the rules, the server will directly output the labeled safety index parameter of the structural stability as the safety index prior parameter of the structural stability safety index. This means that for the key safety index of structural stability, the existing accurate labeled data is used as prior information because it is a reliable result obtained through professional structural analysis.

[0058] For the safety indicator of fire protection level, the server again checks the BIM model safety indicator annotation data of the sample BIM model.

[0059] The safety index parameters of the fire protection level are marked in accordance with the building fire protection code, such as level 1, level 2, etc., which reflects the fire protection capabilities of building materials, fire separations and other aspects.

[0060] Assume that in this sample BIM model, due to some reasons (such as partial loss during data transmission or negligence during original data collection), the safety index parameters of the fire rating are missing.

[0061] At this point, the server will obtain the estimated safety index parameter of the fire rating from the BIM model safety index estimation data of the sample BIM model. This estimated safety index parameter is generated by the BIM model safety index estimation network in the server. The network may estimate a fire rating value, such as level 2, based on multiple factors such as the functional use of the building (office area, relatively high density of people, and high fire protection requirements), the type of building materials (materials that may be used are inferred from other marked parts), etc. through a pre-learned algorithm. Then, this estimated level 2 value is used as the safety index prior parameter of the fire rating safety index.

[0062] For the electrical grounding quality indicator, the server checks the electrical grounding quality safety indicator in the BIM model safety indicator annotation data of the sample BIM model.

[0063] The safety index parameters for marking the electrical grounding quality are reflected through data such as the grounding resistance value obtained by electrical testing equipment.

[0064] Assume that in this sample BIM model, the safety indicator parameters for the electrical grounding are present. For example, the grounding resistance value is accurately marked as 1 ohm, which is an accurate value obtained during the electrical system design and testing process.

[0065] Then, the server will directly use the 1 ohm annotated safety indicator parameter as the safety indicator prior parameter of the electrical grounding safety indicator. This is because accurate annotated data can directly provide reliable prior information without using estimated values.

[0066] For the evacuation channel rationality indicator, when the server checks the evacuation channel rationality safety indicator in the BIM model safety indicator annotation data of the sample BIM model.

[0067] The safety index parameters for the rationality of the evacuation passage may involve the comprehensive evaluation results of multiple factors such as passage width, passage length, location and number of evacuation doors, etc.

[0068] Assume that in this sample BIM model, some design change information was not updated to the annotated data in a timely manner, resulting in the missing annotated safety index parameters for the rationality of the evacuation channel.

[0069] Therefore, the server will obtain the estimated safety index parameter of the rationality of the evacuation channel from the BIM model safety index estimation data. This estimated safety index parameter may be an evaluation value estimated by a machine learning algorithm based on factors such as the total area of ​​the building, the number of people accommodated, and the distribution of different functional areas. For example, the value of the estimated rationality of the evacuation channel is 0.8 (assuming that the value between 0-1 represents the degree of rationality), and this 0.8 is used as the safety index prior parameter of the safety index of the rationality of the evacuation channel.

[0070] For the ventilation safety indicator, the server checks the status of the ventilation safety safety indicator in the BIM model safety indicator annotation data of the sample BIM model.

[0071] The safety index parameters for ventilation safety may be derived based on factors such as the design air volume and air change times of the ventilation system.

[0072] Assume that in this sample BIM model, the ventilation safety indicator parameters are complete. For example, the air change rate is marked as 5 times per hour, which is the value calculated according to the ventilation system design specifications and actual calculations.

[0073] Then, the server will directly use the safety indicator parameter marked 5 times per hour as the safety indicator prior parameter of the ventilation safety safety indicator.

[0074] After determining the safety indicator priori parameters of these Y=5 safety indicators (structural stability, fire protection level, electrical grounding, rationality of evacuation channels, and ventilation safety), the server will generate the BIM model safety indicator priori data of this sample BIM model based on the safety indicator priori parameters of these 5 safety indicators.

[0075] This BIM model safety index prior data will be a set of these five safety index prior parameters, which combines the accurate labeled safety index parameters (when they exist) and the safety index parameters obtained by estimation (when the labeled parameters are missing). This prior data will serve as an important basis for the subsequent network parameter learning of the AI ​​safety optimization network. For example, when training the AI ​​safety optimization network, this prior data will be input into the network together with the sample BIM model. The network will learn and optimize based on this prior information to improve the prediction and optimization capabilities of the BIM model safety indicators, and finally generate a network that can optimize the BIM model data with the target safety identification indicators.

[0076] Therefore, the BIM model safety index prior data of the sample BIM model can be accurately generated, providing a key data foundation for the effective operation of the entire BIM model safety optimization system.

[0077] In a possible implementation, the method further includes: generating a missing indication array corresponding to the partially missing safety identification data, the missing indication array comprising X×Y nodes, each node representing whether a marked safety indicator parameter of a safety indicator of a BIM model is missing.

[0078] In the aforementioned building project, the server has obtained some missing safety identification data of the sample BIM model sequence learned by the AI ​​safety optimization network. Here, it is assumed that there are X=20 BIM model safety indicator annotation data in the missing safety identification data, and each annotation data is for Y=5 safety indicators (structural stability, fire protection level, good electrical grounding, reasonable evacuation channel, ventilation safety).

[0079] The server starts to generate a missing indication array. This missing indication array will have 20×5=100 nodes. For the safety indicator of structural stability of the first BIM model, the server checks whether the annotated safety indicator parameters in the partially missing safety identification data are missing. If so, the corresponding nodes in the missing indication array (the first BIM model corresponds to the 1st to 5th nodes, and the structural stability is the 1st node) will be marked with a specific identifier indicating missing (for example, 1); if there is no missing, it will be marked with an identifier indicating non-missing (for example, 0).

[0080] Next, for the safety indicator of fire rating of the first BIM model, repeat the above inspection process and mark the corresponding node (the second node) in the missing indication array. In this way, all five safety indicators of the first BIM model are inspected and marked in turn.

[0081] Then, the server operates the five safety indicators of the second BIM model in the same way, and accurately marks the nodes corresponding to each safety indicator in the missing indication array. For example, if the safety indicator parameter of the electrical grounding of the second BIM model is missing, the corresponding node (node ​​6+3-1=8) is marked as 1.

[0082] This process continues until all five safety indicators of all 20 BIM models are checked and marked, and finally a complete missing indicator array is formed. This array fully reflects whether the marked safety indicator parameters of each safety indicator of each BIM model in the partially missing safety identification data are missing, providing an intuitive and useful reference for subsequent data processing, analysis, and operations based on the sample BIM model, helping the server to more efficiently manage and process BIM model safety identification-related data.

[0083] In a possible implementation, step S110 includes: Step S111, obtaining complete safety identification data of multiple sample BIM model sequences, wherein the complete safety identification data of each sample BIM model sequence includes a labeled safety indicator parameter of at least one BIM model for at least one safety indicator, and the safety indicators covered by the complete safety identification data of different sample BIM model sequences are all different, and the BIM models covered by the complete safety identification data of different sample BIM model sequences are all different.

[0084] Step S112: Taking each BIM model as a unit, the complete safety identification data of the multiple sample BIM model sequences are merged to generate the partially missing safety identification data.

[0085] In this embodiment, the server first obtains complete safety identification data of a sample BIM model sequence from different design stages of the construction project. For example, in the preliminary design stage of a building, there is a sample BIM model sequence. Each BIM model in this sequence corresponds to a different part of the building, such as the infrastructure part, the main structure part of a first floor, and so on. Each BIM model has annotated safety indicator parameters for specific safety indicators. For the BIM model of the infrastructure part, its safety indicators may include foundation bearing capacity, etc. The annotated safety indicator parameters of the safety indicator of foundation bearing capacity are specific values ​​calculated through geological survey data and structural design, such as being able to withstand a pressure of 50 tons per square meter.

[0086] In the detailed design phase of a building, there is another sample BIM model sequence. The BIM models in this sequence target different safety indicators. For example, for the BIM model of the electrical system inside the building, the safety indicators include the overload protection capacity of the electrical circuit. The safety indicator parameters for the overload protection capacity of the electrical circuit are determined according to the electrical design specifications, such as the overload protection current is set to 20 amperes.

[0087] For example, in the decoration design stage of a building, there is another sample BIM model sequence. The safety indicators targeted by the BIM model in this sequence may be the fire resistance performance of decoration materials, etc. The safety indicator parameters for the fire resistance performance of decoration materials are marked according to the fire resistance material grade standard, such as reaching the B1 fire resistance standard.

[0088] The complete safety identification data of these different sample BIM model sequences cover different BIM models. Each BIM model has its own unique structure and functional characteristics, corresponding to different safety requirements and evaluation standards.

[0089] At the same time, the safety indicators covered by the complete safety identification data of different sample BIM model sequences are also different. This is because different stages and different parts of the building focus on different safety priorities. For example, the preliminary design stage focuses on basic safety indicators such as structural safety, while the decoration design stage focuses more on safety indicators related to decoration materials.

[0090] Next, the server starts to merge the complete safety identification data of these multiple sample BIM model sequences for each BIM model. For example, the server first processes the BIM model data from the preliminary design stage and the detailed design stage. For a BIM model of an infrastructure part that exists in both stages but represents different design states, the server merges the safety indicator data of this BIM model in the two stages.

[0091] Some problems may occur during the merging process. For example, in the preliminary design stage, only the foundation bearing capacity safety indicator may be marked on the BIM model of the basic structure, while in the detailed design stage, in addition to the foundation bearing capacity, the seismic performance of the basic structure is also added as a safety indicator. However, for some reason, during the merging process, the data on the newly added safety indicator of seismic performance may not be completely correctly merged, resulting in some missing data.

[0092] As the server continues to merge the complete safety identification data of more sample BIM model sequences, the data missing situation may become more complicated. For example, when merging the BIM model data of the decoration design stage, some safety indicators related to structure and electricity may be lost during the merging due to differences in data formats at different stages or negligence in data management.

[0093] After merging the complete safety identification data of all relevant sample BIM model sequences, the server finally generated some missing safety identification data. This partially missing safety identification data contains safety indicator annotation data of multiple BIM models, but there is at least one BIM model safety indicator annotation data with missing features, that is, some BIM models have missing annotation safety indicator parameters for some safety indicators.

[0094] Through the above steps, some missing safety identification data for subsequent operations were obtained, which provided a data basis for further optimizing the safety identification and management of BIM models.

[0095] In a possible implementation, step S130 includes: Step S131: Generate an iterative BIM model corresponding to the sample BIM model based on the AI ​​safety optimization network, the sample BIM model and the BIM model safety index prior data of the sample BIM model.

[0096] Step S132: determining a training cost parameter of the AI ​​safety optimization network based on the sample BIM model and the iterative BIM model, wherein the training cost parameter reflects a characteristic distance between the sample BIM model and the iterative BIM model.

[0097] Step S133: Based on the training cost parameters, the network parameters of the AI ​​security optimization network are optimized to generate the AI ​​security optimization network that has completed network parameter learning.

[0098] In this embodiment, in the BIM model security management of the construction project, the AI ​​security optimization network in the server has a specific structure and function. This network includes a first optimization branch and a second optimization branch.

[0099] For a specific sample BIM model, such as the BIM model representing the office area mentioned above, the server first uses AI security to optimize the first optimization branch of the network. This sample BIM model contains structural, fire protection, electrical and other safety information, and already has prior data on BIM model safety indicators generated based on the previous steps.

[0100] The first optimization branch starts to generate a hidden layer expression vector based on the sample BIM model and its BIM model safety index prior data. This hidden layer expression vector is the result of hidden layer neurons calculating and transforming the sample BIM model and its prior data. For example, for the safety index of structural stability, if the safety index prior parameters in the prior data indicate that the structural stability is at a certain level, the hidden layer neurons will convert this information into part of the hidden layer expression vector through complex calculations based on this information and the structural related features in the sample BIM model (such as beam column layout, material strength, etc.).

[0101] Then, the second optimization branch generates an iterative BIM model based on the hidden layer expression vector and the prior data of the BIM model safety index of the sample BIM model. This iterative BIM model is, to a certain extent, an optimization or adjustment result of the sample BIM model. For example, in terms of fire safety, if the prior data of the sample BIM model shows that there is room for improvement in the fire protection level, the iterative BIM model may adjust the fire separation layout inside the building or use materials with higher fire resistance. These adjustments are based on the network's analysis of the prior data and the hidden layer expression vector.

[0102] Next, the server starts to determine the fit evaluation cost parameter in the training cost parameter. Taking the structural stability safety index in the sample BIM model as an example, in the BIM model safety index annotation data of the sample BIM model, the structural stability has accurate annotated safety index parameters (such as structural safety factor), and at the same time, there are also estimated safety index parameters for structural stability in the BIM model safety index estimation data.

[0103] The server determines the fit evaluation cost parameter by comparing these two parameters. If the estimated safety index parameter differs greatly from the annotated safety index parameter, it means that the estimation accuracy is low and the fit evaluation cost parameter will be higher. For example, if the annotated structural safety factor is 1.5 and the estimated safety factor is 1.2, the difference between the two will be quantified as part of the fit evaluation cost parameter. This parameter reflects the estimated performance index of the BIM model safety index estimation data.

[0104] Then determine the first cost parameter, which reflects the explicit feature distance between the iterative BIM model and the sample BIM model. For example, in terms of the explicit feature of the building's external dimensions, if the length, width, and height of the sample BIM model are 50 meters, 30 meters, and 20 meters, respectively, and the length, width, and height of the iterative BIM model become 52 meters, 30 meters, and 20 meters, the server will calculate the first cost parameter based on the difference in these dimensions. The greater the change in dimensions, the greater the value of the first cost parameter, which intuitively reflects the degree of difference between the two models in explicit features such as appearance.

[0105] For the second cost parameter, take the fire rating in the BIM model safety index prior data of the sample BIM model as an example. The server determines it based on the BIM model safety index prior data, the coordinated parameter array of the BIM model safety index prior data, the average data of the BIM model safety index prior data, and the average data of the BIM model safety index estimation data. If the safety index prior parameter of the fire rating is far away from the overall characteristics of the prior data, for example, the prior data as a whole tends to have a higher fire rating, and the prior parameter of this labeled safety index is relatively low, the second cost parameter will reflect this situation.

[0106] For the third cost parameter, consider other safety indicators in the sample BIM model in addition to the annotated safety indicators (such as fire protection level), such as electrical grounding quality. If the prior parameters of the safety indicator of electrical grounding quality have a large characteristic distance compared to the prior data of the BIM model safety indicator, the third cost parameter will reflect this difference. For example, if the prior data indicates that the overall electrical safety is at a high level, but the prior parameters of electrical grounding quality show a large grounding resistance (i.e., poor grounding), the third cost parameter will increase accordingly.

[0107] Next, the server obtains the hidden layer expression vector generated for the sample BIM model from the hidden network layer of the AI ​​security optimization network. This hidden layer expression vector has its own distribution model parameters, including the mean and the morphological vector. The mean reflects the central position of the hidden layer expression vector in the current state distribution, and the morphological vector is related to the distribution shape.

[0108] The server selects a hidden layer expression vector sample set based on the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector, and obtains a pre-built hypothetical distribution model. The hypothetical distribution model represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state.

[0109] Then, the KL divergence is used as a measure of the difference between the distribution model parameters of the hidden layer expression vectors in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model. For example, if the actual distribution of the hidden layer expression vector is significantly different from the assumed distribution model, the KL divergence will be larger. The server determines the relevant parameters for calculating the fourth cost parameter based on the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector, and the covariance matrix related to the mean and morphological vector corresponding to the assumed distribution model. Finally, based on the calculation method and relevant parameters of the KL divergence, the fourth cost parameter is calculated by combining the difference between the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector and the covariance matrix related to the mean and morphological vector corresponding to the assumed distribution model. The calculated fourth cost parameter will be normalized by dividing it by the reference value calculated based on the prior sample BIM model to generate the normalized fourth cost parameter.

[0110] The server determines a probability model optimization cost parameter based on the first cost parameter, the second cost parameter, the third cost parameter and the fourth cost parameter. The probability model optimization cost parameter reflects the characteristic distance between the iterative BIM model generated based on the BIM model safety index prior data and the sample BIM model.

[0111] Finally, the server determines the training cost parameter based on the fit evaluation cost parameter and the probability model optimization cost parameter. This training cost parameter takes into account multiple factors such as estimation accuracy and model feature distance.

[0112] For example, the server optimizes the network parameters of the AI ​​security optimization network based on the calculated training cost parameters. If the training cost parameters are high, it means that the performance of the current AI security optimization network needs to be improved. For example, if the fit evaluation cost parameters are high, it means that the estimation accuracy is poor. The server may adjust the network parameters such as weights and biases in the first optimization branch and the second optimization branch to reduce the fit evaluation cost parameters.

[0113] If the first cost parameter is high, it may mean that the difference in explicit features between the iterative BIM model and the sample BIM model is unreasonable, and the server will adjust the network parameters to reduce this difference. Similarly, if the second, third, and fourth cost parameters are high, the network parameters will be adjusted accordingly, such as adjusting the connection weights between hidden layer neurons.

[0114] Through multiple iterations of optimization, the network parameters are continuously adjusted until the training cost parameters reach a satisfactory level, and finally an AI safety optimization network that has completed network parameter learning is generated. This AI safety optimization network that has completed network parameter learning can better optimize according to the sample BIM model and its prior data, and can be used to optimize the generation of BIM model data with target safety identification indicators.

[0115] In a possible implementation, the training cost parameters include a fitness evaluation cost parameter and a probability model optimization cost parameter, and the probability model optimization cost parameter includes a first cost parameter, a second cost parameter, a third cost parameter, and a fourth cost parameter.

[0116] Step S132 includes: Step S1321, based on the labeled safety indicator parameters of the labeled safety indicators in the BIM model safety indicator labeling data and the estimated safety indicator parameters of the labeled safety indicators in the BIM model safety indicator estimation data, determine the fit evaluation cost parameter, the fit evaluation cost parameter reflects the estimated performance indicator of the BIM model safety indicator estimation data, and the labeled safety indicator is a safety indicator corresponding to the labeled safety indicator parameters that are not missing in the BIM model safety indicator labeling data.

[0117] Among them, when the BIM model safety index estimation data of the sample BIM model is continuously updated and changed as the network parameter information of the AI ​​safety optimization network is learned, the collaborative parameter array and average parameter value of the BIM model safety index prior data are continuously updated and changed accordingly.

[0118] In this embodiment, in the BIM model security management scenario of the construction project, the server processes the relevant data of the sample BIM model. Taking the office area BIM model mentioned above as an example, the model has multiple safety indicators, such as structural stability, fire protection level, good electrical grounding, rationality of evacuation channels, and ventilation safety.

[0119] For the safety index of structural stability, in the safety index annotation data of the BIM model, the annotated safety index parameters are obtained based on detailed structural analysis. Assuming that the structural system of the office area is a frame structure, through the calculation of the bearing capacity of beams and columns, the seismic analysis of the overall structure, etc., the annotated safety index parameter of structural stability is obtained as a safety factor of 1.5. This value is determined after considering multiple factors such as the earthquake intensity of the building area, the design load of the building (including live load and dead load), etc.

[0120] At the same time, the estimated safety index parameters in the BIM model safety index estimation data are generated by a special estimation network. The estimation network may consider factors such as the building's height, number of floors, and type of structural materials, and estimate the safety factor of the structural stability to be 1.3.

[0121] The server uses a specific algorithm to calculate the fit evaluation cost parameter. A common method is the mean square error (MSE). For the structural stability annotation safety indicator, the calculation method is: Mean square error (MSE=(text{annotated safety index parameter}-text{estimated safety index parameter})^2=(1.5-1.3)^2=0.04) For the safety index of fire protection level, the safety index parameter is marked as level 2 according to the building fire protection code. The BIM model safety index estimation data estimates the fire protection level to be 1.8 based on factors such as the density of people in the building and the fire load. Calculate the mean square error: (MSE=(2-1.8)^2=0.04) Assume that for the safety indicator of good electrical grounding, the marked safety indicator parameter is 0.8 ohms of grounding resistance, and the estimated safety indicator parameter is 0.9 ohms. Calculate the mean square error: (MSE=(0.8-0.9)^2=0.01) The server combines the mean square errors calculated for different annotated safety indicators. If we simply take the average, the cost parameter for the fit evaluation is: ((0.04+0.04+0.01) / 3=0.03). This value reflects the estimated performance index of the BIM model safety index estimation data on multiple indicators. If the fit evaluation cost parameter is large, it means that the accuracy of the estimation network needs to be improved.

[0122] Step S1322: determining the first cost parameter based on the iterative BIM model and the sample BIM model, where the first cost parameter reflects the explicit feature distance between the iterative BIM model and the sample BIM model.

[0123] Regarding the explicit feature of building dimensions, the length of the office area in the sample BIM model is 50 meters, the width is 30 meters, and the height is 20 meters. In the iterative BIM model, the length is changed to 51 meters, the width is changed to 30.5 meters, and the height is changed to 20 meters.

[0124] Compute the distance in each dimension: The distance in the length direction is (vert51-50vert=1) meters; The distance in the width direction is (vert30.5-30vert=0.5) meters; The distance in the height direction is (vert20-20vert=0) meters.

[0125] One way for the server to determine the first cost parameter is to simply sum these distances, and the first cost parameter is (1+0.5+0=1.5) meters. This value reflects the distance between the iterative BIM model and the sample BIM model in the explicit feature of the building's external dimensions.

[0126] Let's look at the dominant feature of the evacuation channel layout. In the sample BIM model, there are two evacuation channels. Channel 1 is 2 meters wide and Channel 2 is 1.8 meters wide. The evacuation doors are located at both ends and in the middle of the channels. In the iterative BIM model, the width of Channel 1 is changed to 2.2 meters, the width of Channel 2 is changed to 1.9 meters, and the position of one of the evacuation doors is slightly adjusted.

[0127] For changes in channel width: The distance that channel 1 width changes is (vert2.2-2vert=0.2) meters; The distance that channel 2 width changes is (vert1.9-1.8vert=0.1) meters.

[0128] For the change in the position of the evacuation door, according to the predefined measurement method, it is assumed that the distance corresponding to this change is 0.5 (this value is a quantitative result of the impact of the change in the position of the evacuation door on the evacuation efficiency).

[0129] In summary, this part contributes to the first cost parameter (0.2+0.1+0.5=0.8).

[0130] The server combines the distance contributions of explicit features such as building dimensions and evacuation channel layout to determine the first cost parameter. If there are only these two parts, the first cost parameter is (1.5+0.8=2.3). This first cost parameter fully reflects the distance between the iterative BIM model and the sample BIM model in multiple explicit features.

[0131] Step S1323, based on the BIM model safety indicator prior data, the collaborative parameter array of the BIM model safety indicator prior data, the average data of the BIM model safety indicator prior data and the average data of the BIM model safety indicator estimation data, determine the second cost parameter and the third cost parameter, the second cost parameter reflects the characteristic distance between the safety indicator prior parameters of the marked safety indicator in the BIM model safety indicator prior data and the BIM model safety indicator prior data, and the third cost parameter reflects the characteristic distance between the safety indicator prior parameters of other safety indicators except the marked safety indicator among the Y safety indicators covered by the BIM model safety indicator prior data and the BIM model safety indicator prior data.

[0132] Assume that the fire protection level is used as the annotated safety index. In the BIM model safety index prior data of the sample BIM model, the safety index prior parameter of the fire protection level is level 2.

[0133] The collaborative parameter array of the prior data of the safety indicators of the BIM model contains the relationship information between various safety indicators, such as the degree of correlation between the fire protection level and safety indicators such as building materials and evacuation routes. Assume that the collaborative parameter array indicates that the correlation coefficient between the fire protection level and building materials is 0.8 (indicating a high correlation), and the correlation coefficient with the evacuation route is 0.6.

[0134] The average data of the BIM model safety index prior data reflects the average level of all safety index prior parameters. Assuming that all safety index prior parameters are converted into numerical values ​​(for example, levels 1-3 correspond to 1, 2, and 3 respectively), the average data is 2.2.

[0135] The average data of the estimated data of the BIM model safety index is assumed to be 2.1.

[0136] The server determines the second cost parameter based on these data. One possible calculation method is to consider the difference between the safety index prior parameters of the fire rating and the overall characteristics of the prior data, while considering the association relationship in the collaborative parameter array. For example, if the fire rating is level 2, the difference compared with the average data 2.2 is (vert2-2.2vert=0.2). This difference is then weighted according to the correlation coefficient in the collaborative parameter array. Assuming that the correlation with building materials is high, a weight of 0.8 is given, and the correlation with evacuation channels is the second highest, a weight of 0.6 is given. The calculated second cost parameter is ((0.2*0.8+0.2*0.6) / 2=0.14). This second cost parameter reflects the characteristic distance between the safety index prior parameters of the annotated safety indicator (fire rating) and the BIM model safety index prior data.

[0137] For other safety indicators other than the marked safety indicators (fire protection level), such as electrical grounding quality, in the BIM model safety indicator prior data, the safety indicator prior parameter of electrical grounding quality is the grounding resistance of 0.8 ohms.

[0138] Also consider the coordinated parameter array of the prior data of the BIM model safety index, the average data, and the average data of the estimated data of the BIM model safety index. Assume that the correlation coefficients of electrical grounding and other safety indicators in the coordinated parameter array are different, and the correlation coefficient with structural stability is 0.5, and the correlation coefficient with ventilation safety is 0.4.

[0139] The difference between the prior parameters of the safety index of electrical grounding and the overall characteristics of the prior data is calculated in a similar way to the calculation of the second cost parameter. Assuming that the ideal value of the grounding resistance corresponding to the overall characteristics of the prior data (considering the comprehensive situation of other related safety indicators) is 0.7 ohms, the difference is (vert0.8-0.7vert=0.1).

[0140] By weighting this difference according to the correlation coefficient, the third cost parameter is calculated as ((0.1*0.5+0.1*0.4) / 2=0.045). This third cost parameter reflects the characteristic distance between the safety index prior parameters of other safety indicators (electrical grounding quality) other than the marked safety indicators and the BIM model safety index prior data.

[0141] Step S1324, determining the fourth cost parameter based on the hidden layer expression vector, and the morphological vector and mean of the distribution model parameters followed by the hidden layer expression vector, wherein the fourth cost parameter reflects the discrete parameter value of the hidden layer expression vector compared to the distribution model parameters, and the hidden layer expression vector is the hidden relationship vector of the sample BIM model generated by the hidden network layer of the AI ​​safety optimization network.

[0142] In this embodiment, the server obtains the hidden layer expression vector generated for the sample BIM model from the hidden network layer of the AI ​​security optimization network. This hidden layer expression vector is the result of complex calculation and transformation of the sample BIM model by the hidden layer neurons, and it contains the representation of the sample BIM model in the abstract feature space.

[0143] Assume that the hidden layer expression vector is a multidimensional vector (vec{v}=(v_1,v_2,cdots,v_n)), and the distribution model parameters it follows are the shape vector and the mean. For example, the mean (mu=(mu_1,mu_2,cdots,mu_n)), the shape vector (Sigma) (which can be a parameter representing the distribution shape such as the covariance matrix).

[0144] The server first determines the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector. The mean reflects the central position of the hidden layer expression vector in the current state distribution. For example, (mu_1) may represent the average value on a certain abstract feature dimension. The morphological vector is related to the shape of the distribution, and the elements (Sigma_{ij}) in the covariance matrix represent information such as the correlation between different dimensions.

[0145] The hidden layer expression vector sample set is selected based on the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector. Assume that a set of samples ({vec{v}_1,vec{v}_2,cdots,vec{v}_m}) is selected from multiple calculation results of the hidden layer expression vector.

[0146] Get a pre-built hypothetical distribution model. This hypothetical distribution model represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state, such as a standard normal distribution model (mean is 0, covariance matrix is ​​the unit matrix) or an ideal distribution model determined based on a large amount of prior data.

[0147] KL divergence (Kullback-Leibler / ergence) is used as a measure of the difference between the distribution model parameters of the hidden layer expression vectors in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model.

[0148] According to the calculation formula of KL divergence: (D_{KL}(P||Q)=sum_{i}P(i)logfrac{P(i)}{Q(i)}), where (P) is the actual distribution of the hidden layer expression vector, and (Q) is the assumed distribution model. The server determines the relevant parameters for calculating the fourth cost parameter based on the mean and covariance matrix of the distribution model parameters implementing the hidden layer expression vector and the covariance matrix associated with the mean and morphological vector corresponding to the assumed distribution model.

[0149] For example, assuming that the mean of the actual distribution of the hidden layer expression vector is (mu_p) and the covariance matrix is ​​(Sigma_p), assuming that the mean of the distribution model is (mu_q) and the covariance matrix is ​​(Sigma_q). Calculate the fourth cost parameter (D_{KL}): (D_{KL}=frac{1}{2}(logfrac{vertSigma_qvert}{vertSigma_pvert}-n+tr(Sigma_q^{-1}Sigma_p)+(mu_q-mu_p)^TSigma_q^{-1}(mu_q-mu_p))) The calculated fourth cost parameter may be a specific value, assuming it is 0.3. This value is then divided by the reference value (assuming it is 1) calculated based on the prior sample BIM model for normalization, generating the normalized fourth cost parameter (0.3). This fourth cost parameter reflects the discrete parameter value of the hidden layer expression vector compared to the distribution model parameter.

[0150] Step S1325: Determine the probability model optimization cost parameter based on the first cost parameter, the second cost parameter, the third cost parameter and the fourth cost parameter, wherein the probability model optimization cost parameter reflects the characteristic distance between the iterative BIM model generated based on the BIM model safety index prior data and the sample BIM model.

[0151] Step S1326, determining the training cost parameter based on the fitness evaluation cost parameter and the probability model optimization cost parameter.

[0152] In this embodiment, the server determines the probability model optimization cost parameter based on the first cost parameter, the second cost parameter, the third cost parameter and the fourth cost parameter. For example, assuming that the first cost parameter is 2.3, the second cost parameter is 0.14, the third cost parameter is 0.045, and the fourth cost parameter is 0.3.

[0153] The probability model optimization cost parameter can be a weighted sum or some combination of these cost parameters. Assuming a simple summation method is adopted, the probability model optimization cost parameter is (2.3+0.14+0.045+0.3=2.785). This probability model optimization cost parameter reflects the characteristic distance between the iterative BIM model generated based on the BIM model safety index prior data and the sample BIM model.

[0154] The server determines the training cost parameter based on the fit evaluation cost parameter and the probability model optimization cost parameter. Assume that the fit evaluation cost parameter is 0.03 and the probability model optimization cost parameter is 2.785.

[0155] The training cost parameter can also be the weighted sum of these two cost parameters or some combination. If a simple summation method is used, the training cost parameter is (0.03+2.785=2.815). This training cost parameter takes into account multiple factors such as estimation accuracy (fit evaluation cost parameter) and model feature distance (probabilistic model optimization cost parameter). It will be used to evaluate the performance of the AI ​​security optimization network and guide the optimization process of network parameters.

[0156] During the entire process, when the estimated data of the BIM model safety index of the sample BIM model is continuously updated and changed as the network parameter information of the AI ​​safety optimization network is learned, the coordinated parameter array and average parameter value of the BIM model safety index prior data will be continuously updated and changed. For example, as the network learning progresses, the estimated data estimates each safety index more accurately or changes, which will affect the relationship between each safety index in the prior data (coordinated parameter array) and the overall average level (average parameter value), and then affect the calculation of the second cost parameter and the third cost parameter, and ultimately affect the determination of the entire training cost parameter and the optimization process of the AI ​​safety optimization network.

[0157] In a possible implementation, step S1324 includes: Step S1324-1, obtaining a hidden layer expression vector generated for a sample BIM model from a hidden network layer of the AI ​​safety optimization network, wherein the hidden layer expression vector is a result of hidden layer neuron calculation and conversion of the sample BIM model.

[0158] Step S1324-2, determine the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector, the mean reflects the central position of the hidden layer expression vector in the current state distribution, and the morphological vector is related to the distribution shape.

[0159] Step S1324-3, select a hidden layer expression vector sample set based on the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector, and obtain a pre-constructed hypothetical distribution model, which represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state distribution.

[0160] Step S1324-4, using KL divergence as a measure of the difference between the distribution model parameters of the hidden layer expression vector in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model, based on the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector, and the covariance matrix related to the mean and morphological vector corresponding to the assumed distribution model, determine the relevant parameters for calculating the fourth cost parameter.

[0161] Step S1324-5, based on the calculation method of the KL divergence and the related parameters, the fourth cost parameter is calculated by comprehensively considering the difference between the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector and the covariance matrix related to the mean and morphological vector corresponding to the assumed distribution model.

[0162] Step S1324-6: normalize the calculated fourth cost parameter by dividing it by a reference value calculated based on the prior sample BIM model to generate a normalized fourth cost parameter.

[0163] In this embodiment, in the BIM model safety management system of the construction project, the AI ​​safety optimization network in the server has a complex structure. The hidden network layer of this network contains multiple neurons, which process the data of the sample BIM model. Taking the previously mentioned office area BIM model as an example, the model contains information on structure, fire protection, electricity, etc.

[0164] When the data of this sample BIM model is input into the AI ​​safety optimization network, it first passes through the input layer and then enters the hidden network layer. In the hidden network layer, each neuron calculates the input data according to its preset weight and activation function. For example, for data on structural stability, neurons may calculate based on information such as the connection method of beams and columns, material strength, etc. Assume that there are 100 neurons in the hidden network layer, each neuron receives information from different combinations of the input layer and performs linear combination calculations (such as (z = sum_{i = 1}^{n} w_{i}x_{i}+b), where (w_{i}) is the weight, (x_{i}) is the input data, and (b) is the bias), and then passes through the activation function (such as the ReLU function (y = max(0,z))) for nonlinear transformation.

[0165] After all neurons are calculated and transformed, a hidden layer expression vector for the sample BIM model is generated. This hidden layer expression vector is a high-dimensional vector, assumed to be (vec{v}=(v_1, v_2,cdots, v_{50})) (here the dimension is assumed to be 50), which is the representation of the sample BIM model in the abstract feature space of the hidden network layer. The value of each dimension represents the measurement of the sample BIM model on a certain abstract feature. For example, (v_1) may be related to the abstract representation of the overall stability of the structure in the hidden layer, and (v_2) may be related to a certain hidden feature of fire safety.

[0166] The server starts to determine the mean of the distribution model parameters followed by the hidden layer expression vector. The mean is a parameter that describes the central position of the hidden layer expression vector in the current state distribution. For the obtained hidden layer expression vector (vec{v}=(v_1, v_2,cdots, v_{50})), calculate the mean (mu = (mu_1, mu_2,cdots, mu_{50})).

[0167] The method for calculating the mean is to average the results of multiple calculations of the hidden layer expression vector (assuming there are 1000 calculations). For example, for (mu_1), (mu_1=frac{1}{1000}sum_{j = 1}^{1000}v_{1j}), where (v_{1j}) is the value of the first dimension of the hidden layer expression vector obtained by the (j)th calculation. The same method is used to calculate (mu_2) to (mu_{50}). Assume that the calculated mean is (mu=(0.5, 0.3,cdots, 0.4)). Each element of this mean vector reflects the average value of the hidden layer expression vector in the corresponding dimension, which represents the central position of the hidden layer expression vector in this dimension.

[0168] The morphological vector is related to the distribution shape of the hidden layer expression vector, and the covariance matrix is ​​used as the morphological vector here. The elements (Sigma_{ij}) of the covariance matrix (Sigma) represent the correlation between different dimensions of the hidden layer expression vector.

[0169] The formula for calculating the covariance matrix is ​​(Sigma_{ij}=frac{1}{1000 - 1}sum_{k = 1}^{1000}(v_{ik}-mu_i)(v_{jk}-mu_j)), where (v_{ik}) is the value of the (i)th dimension of the hidden layer expression vector calculated for the (k)th time, and (mu_i) is the (i)th element of the mean vector. This formula is used to calculate a (50times50) covariance matrix (Sigma). This covariance matrix describes the linear relationship between different dimensions of the hidden layer expression vector. For example, if the value of (Sigma_{12}) is large and positive, it means that there is a strong positive correlation between the two dimensions (v_1) and (v_2), that is, when the value of (v_1) is large, the value of (v_2) is also likely to be large.

[0170] Based on the mean and morphological vector (covariance matrix) in the distribution model parameters followed by the hidden layer expression vector, the server selects a hidden layer expression vector sample set. A common method is to construct a multivariate normal distribution (N(mu, Sigma)) based on the mean and covariance matrix, and then randomly sample from this distribution to obtain a sample set. Assume that 100 samples are selected to obtain a hidden layer expression vector sample set ({vec{v}_1, vec{v}_2,cdots, vec{v}_{100}}), each (vec{v}_i) is a vector of the same dimension as the previous hidden layer expression vector (vec{v}). These samples reflect the distribution characteristics of the hidden layer expression vector to a certain extent, and will be used for subsequent comparison with the assumed distribution model.

[0171] The server obtains a pre-built hypothetical distribution model. This hypothetical distribution model represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state. In the example of building safety management, the hypothetical distribution model may be an ideal distribution model obtained based on a large amount of prior building model data or theoretical analysis. For example, assume that it is a multivariate normal distribution (N(mu_0, Sigma_0)) with a mean of (mu_0=(0.4, 0.3,cdots, 0.3)) and a covariance matrix of (Sigma_0) (assuming that (Sigma_0) is a fixed covariance matrix determined based on prior knowledge, representing an ideal relationship between dimensions). This hypothetical distribution model reflects the distribution characteristics that the hidden layer expression vector should have under ideal conditions, such as the distribution of the hidden layer expression vector when the optimal balance is achieved in terms of structural safety, fire safety, etc.

[0172] The server uses KL divergence (Kullback-Leibler / ergence) as a measure of the difference between the distribution model parameters of the hidden layer expression vectors in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model.

[0173] According to the calculation formula of KL divergence: (D_{KL}(P||Q)=sum_{i}P(i)logfrac{P(i)}{Q(i)}), where (P) is the actual distribution of the hidden layer expression vector (reflected by the sample set, whose distribution parameters are the mean (mu) and covariance matrix (Sigma)), and (Q) is the assumed distribution model (whose distribution parameters are the mean (mu_0) and covariance matrix (Sigma_0)).

[0174] In order to calculate KL divergence, it is necessary to determine the relevant parameters for calculating the fourth cost parameter. According to the calculation formula of KL divergence for multivariate normal distribution: (D_{KL}=frac{1}{2}(logfrac{vertSigma_0vert}{vertSigmavert}-n + tr(Sigma_0^{-1}Sigma)+(mu_0 - mu)^TSigma_0^{-1}(mu_0 - mu))), the relevant parameters here include (mu), (Sigma), (mu_0), (Sigma_0) and the operation rules of vectors and matrices.

[0175] For example, (vertSigmavert) is the determinant of the covariance matrix (Sigma), (tr(Sigma_0^{-1}Sigma)) is the trace (i.e., the sum of the main diagonal elements) of the matrix (Sigma_0^{-1}Sigma), and ((mu_0 - mu)^T) is the transpose of the vector (mu_0 - mu). These parameters are based on the actual distribution parameters (mean (mu) and covariance matrix (Sigma)) of the previously determined hidden layer expression vector and the distribution parameters (mean (mu_0) and covariance matrix (Sigma_0)) of the assumed distribution model.

[0176] Based on the calculation method of KL divergence and the related parameters determined previously, the server calculates the fourth cost parameter by combining the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector and the difference between the mean and morphological vector (covariance matrix) corresponding to the assumed distribution model.

[0177] According to the above KL divergence calculation formula for multivariate normal distribution: (D_{KL}=frac{1}{2}(logfrac{vertSigma_0vert}{vertSigmavert}-n + tr(Sigma_0^{-1}Sigma)+(mu_0 - mu)^TSigma_0^{-1}(mu_0 - mu))), substitute the actually calculated (mu=(0.5, 0.3,cdots, 0.4)), (Sigma) (the covariance matrix calculated previously), (mu_0=(0.4, 0.3,cdots, 0.3)), and (Sigma_0) (the assumed covariance matrix) into the formula for calculation.

[0178] Assume that (D_{KL}=0.5) is obtained after calculation (this is just an example result). This value reflects the degree of difference between the actual distribution of the hidden layer expression vector and the assumed distribution model. If the value of (D_{KL}) is large, it means that the distribution of the hidden layer expression vector is quite different from the ideal assumed distribution model, which may mean that there is a certain deviation in the representation of the hidden layer of the AI ​​security optimization network, and the network parameters need to be further adjusted.

[0179] The server divides the calculated fourth cost parameter by a reference value calculated based on the prior sample BIM model to perform normalization processing to generate a normalized fourth cost parameter.

[0180] The method for calculating the reference value based on the prior sample BIM model may be an average KL divergence value or other statistics obtained by similarly calculating multiple prior sample BIM models. Assume that the reference value calculated based on the prior sample BIM model is 1.

[0181] Then the normalized fourth cost parameter is (0.5 / 1 = 0.5). The normalized fourth cost parameter can be more conveniently compared and combined with other cost parameters (such as the first cost parameter, the second cost parameter, the third cost parameter, etc.) to determine the probability model optimization cost parameter and the final training cost parameter, thereby evaluating the performance of the AI ​​security optimization network and guiding the optimization of network parameters.

[0182] In a possible implementation, the AI ​​security optimization network includes a first optimization branch and a second optimization branch.

[0183] The first optimization branch is used to generate a hidden layer expression vector based on the sample BIM model and the BIM model safety index prior data of the sample BIM model. The average data and morphological vector of the distribution model parameters followed by the hidden layer expression vector are determined by the network parameter information of the first optimization branch, the sample BIM model and the BIM model safety index prior data of the sample BIM model.

[0184] The second optimization branch is used to generate the iterative BIM model based on the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model. The average data and morphological vector of the distribution model parameters followed by the iterative BIM model are determined by the network parameter information of the second optimization branch, the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model.

[0185] In this embodiment, in the BIM model security management scenario of the construction project, the AI ​​security optimization network in the server starts working. For the office area sample BIM model mentioned earlier, the office area sample BIM model contains information related to various safety indicators, such as structural stability, fire protection level, good electrical grounding, etc. At the same time, there is already BIM model safety indicator prior data for this sample BIM model, which combines accurate annotated safety indicator parameters (when they exist) and estimated safety indicator parameters (when annotated parameters are missing).

[0186] The first optimization branch receives the sample BIM model and its BIM model safety index prior data as input. These input data contain rich information. For example, in terms of structural stability, the safety index prior parameters in the prior data may include key structural component dimensions, material strength and other information; in terms of fire rating, it may include the fire performance level of building materials, the setting of fire separation and other information.

[0187] There are a series of hidden layer neurons inside the first optimization branch. These neurons start to process the input data. Taking the data related to structural stability as an example, the neurons will perform linear combination calculations according to the preset weights based on the information such as component size and material strength in the prior data. Assume that for a neuron related to structural stability, its calculation method is (z = w_1times text{component size}+ w_2times text{material strength}+b), where (w_1) and (w_2) are weights and (b) is a bias. This calculation result (z) is then transformed nonlinearly through an activation function (such as the Sigmoid function (y=frac{1}{1 +e^{-z}})) to obtain an intermediate result.

[0188] For data related to fire rating, neurons will also perform similar calculations. For example, according to information such as the fire performance level of building materials and the setting of fire separation, the corresponding intermediate results are calculated through different weight combinations and activation functions. This process will be carried out in parallel in multiple hidden layer neurons of the first optimization branch, and each neuron focuses on different parts of the input data or different feature combinations.

[0189] After calculation and transformation by hidden layer neurons, the first optimization branch generates a hidden layer expression vector. This hidden layer expression vector is a multi-dimensional vector, such as (vec{v}=(v_1, v_2,cdots, v_{n})), where the value of each dimension represents the measurement of the sample BIM model on a certain abstract feature. These abstract features are obtained by complex calculation and transformation of the original input data (sample BIM model and prior data) by hidden layer neurons.

[0190] The average data (mean) and morphological vector (here taking the covariance matrix as an example) of the distribution model parameters followed by the hidden layer expression vector are determined by the network parameter information of the first optimization branch, the sample BIM model, and the BIM model safety index prior data of the sample BIM model.

[0191] The process of calculating the mean is as follows: for multiple calculations of the hidden layer expression vector (assuming (m) calculations are performed), calculate the average value of each dimension. For example, for the (v_1) dimension, the mean (mu_1=frac{1}{m}sum_{i = 1}^{m}v_{1i}), where (v_{1i}) is the value of the first dimension of the hidden layer expression vector obtained by the (i)th calculation. The entire mean vector (mu = (mu_1, mu_2,cdots, mu_{n})) is obtained through similar calculations.

[0192] When calculating the covariance matrix (morphological vector), the formula (Sigma_{ij}=frac{1}{m - 1}sum_{k =1}^{m}(v_{ik}-mu_i)(v_{jk}-mu_j)) is used, where (Sigma_{ij}) is the (i)th row and (j)th column element of the covariance matrix, (v_{ik}) is the value of the (i)th dimension of the hidden layer expression vector calculated for the (k)th time, and (mu_i) is the (i)th element of the mean vector. This calculation yields the covariance matrix (Sigma), which describes the correlation between different dimensions of the hidden layer expression vector. These distribution model parameters (mean and covariance matrix) reflect the distribution characteristics of the hidden layer expression vector in the first optimization branch, and will affect the subsequent operation of the second optimization branch and the performance evaluation of the entire AI security optimization network.

[0193] The second optimization branch receives as input the hidden layer expression vector from the first optimization branch and the BIM model safety index prior data of the sample BIM model. The hidden layer expression vector contains the representation of the sample BIM model in the abstract feature space, while the prior data provides important information about the safety index of the sample BIM model.

[0194] The second optimization branch also has its own network structure, including multiple neuron layers, which will generate iterative BIM models based on the input data.

[0195] For each dimension value in the hidden layer expression vector, the neurons of the second optimization branch will calculate according to the preset weights and the corresponding safety indicator prior parameters in the prior data. For example, for the hidden layer expression vector dimension related to structural stability and the prior data of structural stability, the neuron may perform the following calculation: (z' = w'_1timesv_{s}+ w'_2times text{structural stability prior parameter}+b'), where (w'_1) and (w'_2) are weights, (v_{s}) is the dimension value related to structural stability in the hidden layer expression vector, and (b') is the bias. This calculation result (z') is nonlinearly transformed by an activation function (such as the ReLU function (y' = max(0,z'))) to obtain an intermediate result.

[0196] Similar calculations are performed for other safety indicators, such as fire rating, electrical grounding, etc. These calculations are performed step by step in multiple neuron layers in the second optimization branch, continuously transforming and integrating the input data.

[0197] After a series of neural calculations, the second optimization branch generates an iterative BIM model. This iterative BIM model is an optimization or adjustment result of the sample BIM model. For example, in terms of structure, the iterative BIM model may adjust the size or material selection of certain components; in terms of fire protection, it may change the layout of fire separation or use different fire protection materials.

[0198] The average data (mean) and morphological vector (covariance matrix) of the distribution model parameters followed by the iterative BIM model are determined by the network parameter information of the second optimization branch, the hidden layer expression vector, and the BIM model safety index prior data of the sample BIM model.

[0199] The process of determining the mean is similar to the first optimization branch, but based on the calculation results of the second optimization branch. Assume that a feature of the iterative BIM model is calculated (p) times and the result is (x_1, x_2,cdots, x_{p}). For this feature, the mean is (mu'=frac{1}{p}sum_{i = 1}^{p}x_{i}). This calculation is performed for multiple features of the iterative BIM model to obtain the mean vector.

[0200] When calculating the covariance matrix, a similar formula is used to calculate the covariance matrix based on the relationship between the calculation results of different features of the iterative BIM model. This covariance matrix describes the correlation between the features of the iterative BIM model and reflects the distribution characteristics of the iterative BIM model. These distribution model parameters are important for evaluating the relationship between the iterative BIM model and the sample BIM model, as well as the performance of the entire AI safety optimization network.

[0201] Through the collaborative work of the first optimization branch and the second optimization branch, the AI ​​safety optimization network can generate an iterative BIM model based on the sample BIM model and its prior data, and by determining the hidden layer expression vector and the iterative BIM model distribution model parameters, it provides an important basis for subsequent network parameter learning and model optimization, so as to improve the performance and optimization capabilities of the BIM model in terms of safety.

[0202] In a possible implementation manner, the BIM model safety index estimation data of the sample BIM model is generated by a BIM model safety index estimation network, and the method further includes: A. When the number of BIM models in the partially missing safety identification data is greater than a threshold number, the BIM model safety index estimation network calls a first deep learning network that has not yet performed network parameter pre-learning, and, during the network parameter learning process of the AI ​​safety optimization network, optimizes the network parameters of the first deep learning network based on the loss function value of the BIM model safety index annotation data of the sample BIM model and the BIM model safety index estimation data of the sample BIM model.

[0203] Alternatively, B. when the number of BIM models in the partially missing safety identification data is not greater than a threshold number, the BIM model safety index estimation network calls a second deep learning network after network parameter pre-learning, and the second deep learning network is an artificial intelligence network with BIM model safety index estimation performance generated after network parameter pre-learning using the complete safety identification data of a sample BIM model sequence, and the complete safety identification data includes a variety of sample BIM models and corresponding safety index parameters.

[0204] In this embodiment, in the BIM model safety management system of the construction project, the server is responsible for the management and data processing of the entire process. Still taking the previously mentioned office area sample BIM model containing multiple safety indicators (such as structural stability, fire protection level, good electrical grounding, etc.) as an example.

[0205] For each sample BIM model, there is some missing safety identification data. These data contain multiple BIM model safety indicator annotation data, some of which are missing.

[0206] Assume that the server sets a threshold number, for example, 50. When the number of BIM models in the partially missing safety identification data is greater than the threshold number, the server adopts a specific strategy.

[0207] Assume that in the current construction project, the number of BIM models in the partially missing safety identification data is 60, which is greater than 50. At this time, the BIM model safety index estimation network will call the first deep learning network that has not yet performed network parameter pre-learning.

[0208] This first deep learning network is an initial state network, which has not been pre-learned and optimized for the BIM model safety index estimation task. It may have some default network structures, such as containing multiple hidden layers, each layer has a certain number of neurons (such as 3 hidden layers, each layer has 100, 50, 20 neurons respectively), and uses the default weight initialization method (such as random initialization).

[0209] When the first deep learning network is called, it starts to estimate the BIM model safety index of the sample BIM model. Taking the structural stability index as an example, the first deep learning network will receive relevant feature data of the sample BIM model, such as the geometric shape of the building, the type of structural material, etc. as input.

[0210] The neurons in the network perform calculations based on the input data. In the input layer, the data is distributed according to the number of neurons. For example, if the input data is a vector (vec{x}=(x_1, x_2,cdots, x_{n})), and the input layer has 100 neurons, then each neuron will receive a part of the input data or some kind of transformed input data.

[0211] Then, the data is propagated and calculated in the hidden layer. For the first hidden layer, assume that the calculation method of the neuron is (z_{ij} = sum_{k = 1}^{n} w_{ijk}x_{k}+b_{ij}), where (w_{ijk}) is the weight of the (i)th neuron in the (j)th layer, (x_{k}) is the (k)th element of the input data, and (b_{ij}) is the bias of the (i)th neuron in the (j)th layer. Then, it is transformed nonlinearly through an activation function (such as the ReLU function (y_{ij} = max(0,z_{ij}))).

[0212] After calculations in multiple hidden layers, the estimated safety index parameters of the structural stability are obtained at the output layer. This estimation process is performed for each safety index of the sample BIM model.

[0213] During the network parameter learning process of the AI ​​security optimization network, the server will optimize the network parameters of the first deep learning network based on the loss function value of the BIM model safety indicator annotation data of the sample BIM model and the BIM model safety indicator estimation data of the sample BIM model.

[0214] For example, using mean square error (MSE) as the loss function, for the structural stability index, if the labeled safety index parameter is (s) and the estimated safety index parameter is (hat{s}), then the loss function value is (L = (s - hat{s})^2). For all safety indicators, calculate the total loss function value.

[0215] Then, according to the loss function value, the weights and biases of the first deep learning network are updated using an optimization algorithm (such as the stochastic gradient descent algorithm). In the stochastic gradient descent algorithm, the update formulas for weights and biases are (w_{ijk}=w_{ijk}-alphafrac{partial L}{partial w_{ijk}}) and (b_{ij}=b_{ij}-alphafrac{partial L}{partial b_{ij}}), where (alpha) is the learning rate, which determines the step size of each update. By continuously iterating this process, the network parameters of the first deep learning network are optimized, thereby improving the estimation performance of the safety indicators of the BIM model.

[0216] Assume that in another construction project or at different stages of the same project, the number of BIM models in the partially missing safety identification data is 40 and not more than 50. At this time, the BIM model safety index estimation network will call the second deep learning network after pre-learning of network parameters.

[0217] This second deep learning network is an artificial intelligence network with BIM model safety index estimation performance generated by pre-learning network parameters using the complete safety identification data of the sample BIM model sequence. This complete safety identification data includes a variety of sample BIM models and corresponding safety index parameters. For example, this complete safety identification data may contain sample BIM models from different building areas (such as office areas, commercial areas, residential areas, etc.), and each sample BIM model has accurate labeled parameters for safety indicators such as structural stability, fire protection level, and electrical grounding quality.

[0218] During the pre-learning process, the second deep learning network has learned the relationship patterns between different sample BIM models and safety indicators. When it is called to estimate the safety indicators of the current sample BIM model, it can estimate more accurately based on the characteristics of the sample BIM model. For example, for a new office area sample BIM model, the second deep learning network can quickly and accurately estimate the estimated safety indicator parameters of each safety indicator based on the structural characteristics and safety indicator relationships of the previously learned office area sample BIM model, without the need for a learning and optimization process from scratch like the first deep learning network.

[0219] Through such a mechanism, different deep learning networks are selected to estimate the safety indicators of BIM models according to the relationship between the number of BIM models and the number of thresholds in some missing safety identification data, and network parameters are optimized when necessary. This can improve the efficiency and accuracy of BIM model safety indicator estimation, thereby providing better support for the operation of the entire AI safety optimization network and the safety management of BIM models.

[0220] Figure 2 The hardware structure of the smart construction site service system 100 provided in the embodiment of the present application for implementing the above-mentioned AI safety identification method based on BIM and smart construction sites is shown, as Figure 2 As shown, the smart construction site service system 100 may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0221] In one possible design, the smart construction site service system 100 can be a single server or a server group. The server group can be centralized or distributed (for example, the smart construction site service system 100 can be a distributed system). In some embodiments, the smart construction site service system 100 can be local or remote. For example, the smart construction site service system 100 can access information and / or data stored in a machine-readable storage medium 120 via a network. For another example, the smart construction site service system 100 can be directly connected to the machine-readable storage medium 120 to access the stored information and / or data. In some embodiments, the smart construction site service system 100 can be implemented on a smart construction site service system. By way of example only, the smart construction site service system may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc. or any aggregation thereof.

[0222] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions used by the smart construction site service system 100 to execute or use to complete the exemplary methods described in this application.

[0223] During the specific implementation process, one or more processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the AI ​​safety identification method based on BIM and smart construction sites in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0224] The specific implementation process of the processor 110 can be found in the various method embodiments executed by the above-mentioned smart construction site service system 100. The implementation principles and technical effects are similar and will not be repeated in this embodiment.

[0225] In addition, an embodiment of the present application also provides a readable storage medium, in which computer executable instructions are set. When the processor executes the computer executable instructions, the above-mentioned AI safety identification method based on BIM and smart construction sites is implemented.

[0226] It should be noted that in order to simplify the description disclosed in this application and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, multiple features are sometimes combined into one embodiment, drawings, or descriptions thereof. Similarly, it should be noted that in order to simplify the description disclosed in this application and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, multiple features are sometimes combined into one embodiment, drawings, or descriptions thereof.

Claims

1. An AI safety identification method based on BIM and smart construction sites, characterized in that: The method comprises: Obtaining partially missing safety identification data of a sample BIM model sequence learned by an AI safety optimization network, wherein the partially missing safety identification data includes X BIM model safety indicator annotation data, each BIM model safety indicator annotation data includes annotated safety indicator parameters of a BIM model for Y safety indicators, and at least one BIM model safety indicator annotation data with missing features exists among the X BIM model safety indicator annotation data, and at least one labeled safety indicator parameter of a safety indicator is missing among the labeled safety indicator parameters of the Y safety indicators included in the BIM model safety indicator annotation data with missing features; For each sample BIM model, if the BIM model safety index annotation data corresponding to the sample BIM model in the partially missing safety identification data is feature-missing, then based on the BIM model safety index estimation data and the BIM model safety index annotation data of the sample BIM model, the BIM model safety index prior data of the sample BIM model is generated, and the BIM model safety index estimation data of the sample BIM model includes: estimated safety index parameters of the sample BIM model for at least one safety index generated by estimation; Based on the BIM model safety index prior data of the sample BIM model, network parameter learning is performed on the AI ​​safety optimization network to generate an AI safety optimization network that has completed network parameter learning. The AI ​​safety optimization network that has completed network parameter learning is used to optimize and generate BIM model data with target safety identification indicators.

2. The AI ​​safety identification method based on BIM and smart construction site according to claim 1 is characterized in that: The step of generating BIM model safety index prior data of the sample BIM model based on the BIM model safety index estimation data and the BIM model safety index annotation data of the sample BIM model includes: For each of the Y safety indicators, if the annotated safety indicator parameter of the safety indicator covered in the BIM model safety indicator annotated data of the sample BIM model is missing, obtaining the estimated safety indicator parameter of the safety indicator from the BIM model safety indicator estimation data of the sample BIM model as the safety indicator priori parameter of the safety indicator; If there is no missing annotated safety indicator parameter of the safety indicator covered in the BIM model safety indicator annotated data of the sample BIM model, outputting the annotated safety indicator parameter of the safety indicator as a safety indicator priori parameter of the safety indicator; Based on the safety index prior parameters of the Y safety indexes, BIM model safety index prior data of the sample BIM model is generated.

3. The AI ​​safety identification method based on BIM and smart construction site according to claim 2 is characterized in that: The method further comprises: A missing indication array corresponding to the partially missing safety identification data is generated, wherein the missing indication array includes X×Y nodes, and each node represents whether a marked safety indicator parameter of a safety indicator of a BIM model is missing.

4. The AI ​​safety identification method based on BIM and smart construction site according to claim 1 is characterized in that: The obtaining of partially missing safety identification data of the sample BIM model sequence learned by the AI ​​safety optimization network includes: Acquire complete safety identification data of multiple sample BIM model sequences, wherein the complete safety identification data of each sample BIM model sequence includes a labeled safety indicator parameter of at least one BIM model for at least one safety indicator, and the safety indicators covered by the complete safety identification data of different sample BIM model sequences are all different, and the BIM models covered by the complete safety identification data of different sample BIM model sequences are all different; Taking each BIM model as a unit, the complete safety identification data of the multiple sample BIM model sequences are merged to generate the partially missing safety identification data.

5. The AI ​​safety identification method based on BIM and smart construction site according to claim 1 is characterized in that: The method of performing network parameter learning on the AI ​​safety optimization network based on the BIM model safety index prior data of the sample BIM model to generate an AI safety optimization network that has completed network parameter learning includes: Generate an iterative BIM model corresponding to the sample BIM model based on the AI ​​safety optimization network, based on the sample BIM model and the BIM model safety index prior data of the sample BIM model; Based on the sample BIM model and the iterative BIM model, determining a training cost parameter of the AI ​​safety optimization network, wherein the training cost parameter reflects a characteristic distance between the sample BIM model and the iterative BIM model; Based on the training cost parameters, the network parameters of the AI ​​security optimization network are optimized to generate the AI ​​security optimization network that has completed network parameter learning.

6. The AI ​​safety identification method based on BIM and smart construction site according to claim 5 is characterized in that: The training cost parameters include a fitness evaluation cost parameter and a probability model optimization cost parameter, and the probability model optimization cost parameter includes a first cost parameter, a second cost parameter, a third cost parameter and a fourth cost parameter; The determining, based on the sample BIM model and the iterative BIM model, a training cost parameter of the AI ​​safety optimization network includes: Determine the goodness of fit evaluation cost parameter based on the labeled safety indicator parameters of the labeled safety indicators in the BIM model safety indicator labeled data and the estimated safety indicator parameters of the labeled safety indicators in the BIM model safety indicator estimation data, wherein the goodness of fit evaluation cost parameter reflects the estimated performance indicator of the BIM model safety indicator estimation data, and the labeled safety indicator is a safety indicator corresponding to the labeled safety indicator parameters that are not missing in the BIM model safety indicator labeled data; Determine the first cost parameter based on the iterative BIM model and the sample BIM model, where the first cost parameter reflects the explicit feature distance between the iterative BIM model and the sample BIM model; Based on the BIM model safety indicator prior data, the collaborative parameter array of the BIM model safety indicator prior data, the average data of the BIM model safety indicator prior data, and the average data of the BIM model safety indicator estimation data, the second cost parameter and the third cost parameter are determined, the second cost parameter reflecting the characteristic distance between the safety indicator prior parameters of the annotated safety indicator in the BIM model safety indicator prior data and the BIM model safety indicator prior data, and the third cost parameter reflecting the characteristic distance between the safety indicator prior parameters of other safety indicators except the annotated safety indicator among the Y safety indicators covered by the BIM model safety indicator prior data and the BIM model safety indicator prior data; Determine the fourth cost parameter based on the hidden layer expression vector, and the morphological vector and the mean of the distribution model parameter followed by the hidden layer expression vector, wherein the fourth cost parameter reflects the discrete parameter value of the hidden layer expression vector compared to the distribution model parameter, and the hidden layer expression vector is a hidden relationship vector of the sample BIM model generated by the hidden network layer of the AI ​​safety optimization network; Based on the first cost parameter, the second cost parameter, the third cost parameter and the fourth cost parameter, determining the probability model optimization cost parameter, wherein the probability model optimization cost parameter reflects the characteristic distance between the iterative BIM model generated based on the BIM model safety index prior data and the sample BIM model; Determining the training cost parameter based on the fitness evaluation cost parameter and the probability model optimization cost parameter; Among them, when the BIM model safety index estimation data of the sample BIM model is continuously updated and changed as the network parameter information of the AI ​​safety optimization network is learned, the collaborative parameter array and average parameter value of the BIM model safety index prior data are continuously updated and changed accordingly.

7. The AI ​​safety identification method based on BIM and smart construction site according to claim 6 is characterized in that: The step of determining the fourth cost parameter based on the hidden layer expression vector, and the morphological vector and mean value of the distribution model parameter followed by the hidden layer expression vector comprises: Obtaining a hidden layer expression vector generated for the sample BIM model from the hidden network layer of the AI ​​safety optimization network, wherein the hidden layer expression vector is a result of hidden layer neuron calculation and conversion of the sample BIM model; Determine a mean value and a morphological vector in the distribution model parameters followed by the hidden layer expression vector, wherein the mean value reflects the central position of the hidden layer expression vector in the current state distribution, and the morphological vector is related to the distribution shape; Selecting a hidden layer expression vector sample set based on the mean and morphological vector in the distribution model parameters followed by the hidden layer expression vector, and obtaining a pre-constructed hypothetical distribution model, wherein the hypothetical distribution model represents the distribution pattern that the hidden layer expression vector should follow under the hypothetical state distribution; Using KL divergence as a measure of the difference between the distribution model parameters of the hidden layer expression vectors in the hidden layer expression vector sample set and the distribution model parameters of the assumed distribution model, determining the relevant parameters for calculating the fourth cost parameter according to the mean and covariance matrix of the distribution model parameters of the hidden layer expression vectors and the covariance matrix associated with the mean and morphological vector corresponding to the assumed distribution model; Based on the calculation method of the KL divergence and the related parameters, the fourth cost parameter is calculated by combining the mean and covariance matrix of the distribution model parameters of the hidden layer expression vector and the difference between the mean and covariance matrix related to the morphological vector corresponding to the assumed distribution model; The calculated fourth cost parameter is divided by a reference value calculated based on the prior sample BIM model for normalization to generate a normalized fourth cost parameter.

8. The AI ​​safety identification method based on BIM and smart construction site according to claim 5 is characterized in that: The AI ​​security optimization network includes a first optimization branch and a second optimization branch; The first optimization branch is used to generate a hidden layer expression vector based on the sample BIM model and the BIM model safety index prior data of the sample BIM model, wherein the average data and the morphological vector of the distribution model parameter followed by the hidden layer expression vector are determined by the network parameter information of the first optimization branch, the sample BIM model and the BIM model safety index prior data of the sample BIM model; The second optimization branch is used to generate the iterative BIM model based on the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model. The average data and morphological vector of the distribution model parameters followed by the iterative BIM model are determined by the network parameter information of the second optimization branch, the hidden layer expression vector and the BIM model safety index prior data of the sample BIM model.

9. The AI ​​safety identification method based on BIM and smart construction site according to claim 1 is characterized in that: The BIM model safety index estimation data of the sample BIM model is generated by a BIM model safety index estimation network, and the method further includes: When the number of BIM models in the partially missing safety identification data is greater than a threshold number, the BIM model safety index estimation network calls a first deep learning network that has not yet performed network parameter pre-learning, and in the network parameter learning process of the AI ​​safety optimization network, the network parameters of the first deep learning network are optimized according to the loss function value of the BIM model safety index annotation data of the sample BIM model and the BIM model safety index estimation data of the sample BIM model; Alternatively, when the number of BIM models in the partially missing safety identification data is not greater than a threshold number, the BIM model safety index estimation network calls a second deep learning network after network parameter pre-learning, and the second deep learning network is an artificial intelligence network with BIM model safety index estimation performance generated after network parameter pre-learning using the complete safety identification data of a sample BIM model sequence, and the complete safety identification data includes a variety of sample BIM models and corresponding safety index parameters.

10. A smart construction site service system, characterized in that: The smart construction site service system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the AI ​​safety identification method based on BIM and smart construction site as described in any one of claims 1 to 9 above.

Images